[HN Gopher] Unsolved Problems in ML Safety
___________________________________________________________________
Unsolved Problems in ML Safety
Author : pramodbiligiri
Score : 56 points
Date : 2021-10-09 12:22 UTC (2 days ago)
(HTM) web link (arxiv.org)
(TXT) w3m dump (arxiv.org)
| potiuper wrote:
| ML=machine learning; not meta language.
| unholiness wrote:
| Some overlap here with the 2017 paper from Google Brain et al:
| Concrete Problems in AI Safety:
| https://arxiv.org/pdf/1606.06565.pdf
|
| A nice video from the wonderful AI safety communicator Robert
| Miles outlines those problems:
| https://www.youtube.com/watch?v=AjyM-f8rDpg
| sgt101 wrote:
| These are unsolved in the sense that there is no formal or
| comprehensive solution to them, and in the limit that means that
| ML cannot be used for some applications.
|
| On the other hand there are mitigations that can be implemented
| to increase the confidence that we have in ML centric solutions -
| and that can bring some applications back into scope.
|
| Figure 5 shows this using the metaphor of layers of swiss cheese
| that other folk have used with relation to stopping Covid. Of
| course we can't stop Covid completely, but by being careful and
| getting a vaccine, ventilation and masks you can improve your
| chances.
___________________________________________________________________
(page generated 2021-10-11 23:02 UTC)